In-Situ Gravimetric Studies of Wetting, Penetration and Wear ......magnesia spinel brick (added...

of 18

In-Situ Gravimetric Studies of Wetting, Penetration andWear of Refractories by Molten Slags

Dongsheng Xie, Ty Tran, and Sharif Jahanshahi

G K Williams Cooperative Research Centre for Extractive MetallurgyCSIRO Division of Minerals, Box 312, Clayton South, 3169, Australia

ABSTRACT

Interactions between refractories and slags could involve wetting, penetration, dissolution andstructural failure of refractories. Conventional refractory testing techniques, relying on postmortem analysis, provide limited information on the interactive process. To overcome thislimitation, a new gravimetric technique, based on wetting and capillary effects, has beendeveloped to provide direct information on dynamic processes of wetting,penetration/dissolution, and wear of the refractories by molten slags.

The technique has been successfully applied to three slag-refractory systems pertinent to typicalferrous and non-ferrous processes. The gravimetric results revealed that the interactions betweenrefractories and slags varied from one system to another, from simultaneous penetration anddissolution, predominant penetration with little dissolution, to severe penetration/cracking andrefractory failure due to slag attack.

Characteristics of the wetting and penetration/dissolution were analysed from the gravimetricdata. Some important interfacial properties, such as wettability (defined as γ⋅cosθ, where γ issurface tension and θ the contact angle), and the amount of slag penetration or the net weightchange due to penetration and dissolution were estimated. The slag penetration is generally non-uniform and the rate of penetration is influenced by interfacial properties, slag properties andstructural characteristics of the refractories. Iron oxide in the slag had a significant effect onwetting and penetration of magnesia based refractories because of its effect on wettability andkinematic viscosity of the slag and its preferential reactions with refractory grains.

of 18

I. INTRODUCTION

Refractory wear due to slag attack is a major concern for many metallurgical processes.[1,2]Typical refractory-slag interactions involve sequential steps of wetting, slag penetration,refractory dissolution and structural failure. A good knowledge of these processes is importantfor our understanding of the mechanisms of refractory wear. Slag attack has been commonlystudied[1,3-6] by the cup (crucible) test, dip (finger) test, rotary slag test, and induction test. Theseverity of the slag attack are determined by post-mortem sample analysis, which is inherentlytedious and often subjected to human error and interpretations.

To overcome this limitation, a new gravimetric technique has been developed to study the kineticprocess of slag-refractory interactions. Based on the principal of wetting and capillary flow, thein-situ gravimetric measurements provide direct information on dynamic wetting, penetration anddissolution of refractories by molten slags. The new technique has been successfully applied toseveral refractory-slag systems pertinent to typical ferrous and non-ferrous processes.Characteristics of wetting and slag penetration into porous refractories and effect of iron oxidecontent in the slag were investigated.

II. EXPERIMENTAL

The experimental set-up is schematically shown in Figure 1. The refractory specimen wassuspended above a slag bath and the bottom of the sample made to just contact the surface of theslag contained in a platinum crucible. The weight changes recorded by an electronic balance werelogged to a personal computer. The facility was gas-tight to allow control of experimentalatmosphere. Argon or other gases were introduced through the gas-enclosed balance box aboveand passed through a sealed joining tube to enter the furnace. This also served to protect thebalance box from excessive heat. Before use, gases were purified to remove traces of moistureand oxygen by passing through a column of silica gel and the copper turning held at 500°C.

The slag was contained in a platinum (10 wt% rhodium) crucible, which sat on an aluminacastable platform. The position of the crucible could be adjusted horizontally to allow accuratecentral alignment of the hanging refractory sample. The crucible and platform were supported byan alumina tube, which passed through a gas-tight hole in the bottom furnace end cap and wassecurely mounted on a horizontal support arm extending from a linear actuator. The verticalposition of the crucible, which was crucial to the weight measurement, was accurately controlledto within ± 0.1 mm by the linear actuator. Pre-melt synthetic slags of 150-180 g were chargedinto the crucible (about 70 mm diameter) and the depth of molten slag bath was approximately13-16 mm.

The cylindrical refractory samples (Φ20-40 mm) were core drilled from bricks. A small hole wasdrilled at the top of the sample, into which a small diameter alumina tube was cemented coaxiallywith the cylinder sample using a specially designed alignment tool. Platinum wire was used toconnect the small alumina tube to the balance hook. The output of the balance was transferred toa personal computer, and the data could be logged at speeds as high as 5 readings per second to

of 18

give sufficient data points for analysis, particularly during the initial contact period. The accuracyof the balance was 0.1 mg.

The procedure of the experiments was as follows. After the refractory sample was carefullyaligned and positioned within 10-20 mm from the surface of the pre-melted slag held in thecrucible, the furnace and its contents were heated to desired temperature under a controlled gasatmosphere (eg argon). The logging of the balance reading was then initiated and the cruciblewas raised slowly until contact with the refractory sample was made. This was observed by asharp increase in the balance weight reading (this applied to all systems where the slag wetted therefractory samples). The weight changes were continuously recorded over the desired period ofthe slag-refractory contact, which varied from about 40 min to 2 hours depending on the system.At the end of the gravimetric measurements, the crucible was lowered until the slag contact withthe refractory sample was ceased. Immediately after this, the refractory sample was raised about200 mm into the cooler zone of the furnace, in order to freeze the slag soaked inside the sample.After the furnace was cooled to room temperature, the samples were removed and photographed.The samples were then sectioned and analysed for the chemical composition using XRF andexamined using SEM or microprobe techniques.

III. RESULTSThree slag-refractory systems were studied. In the first case, interactions between a silicate slagin the system of CaO-SiO2-Al2O3-MgO and a magnesia spinel brick were studied and effect ofrefractory sample size and addition of 5 wt% FeOx in the slag were investigated. Studies on asecond model system investigated attack by a calcium ferrite type slag on a magnesite and amagchrome brick. In the last case, attack by a ladle type slag on four alumina castablerefractories was investigated. The experimental results are described in details below.

A. Silicate Slag and Magnesia Spinel Refractory

The chemical compositions of the refractory and slag are listed in Table 1. The refractory was amagnesia spinel brick (added MgO⋅Al2O3 spinel shown in Table 1 as alumina) and threecylindrical samples were used with diameters of 20, 30 and 40 mm. About 150 g of the syntheticslags were used for each experiment. The experiments were conducted at 1500 °C under argonand the slag-refractory contact time was about 40 minutes. The gravimetric curves obtained andthe photographs of the refractory samples after the tests are shown in Figure 2. The resultsshowed that the silicate slag penetrated extensively into the refractory and, simultaneously, therefractory was dissolved into the slag, particularly at the slag contact level (showing typicalnecking and slag line corrosion due to the Marangoni effect).[7,8] The bottom of the sample wasalso attacked severely and showed “crater” shaped corrosion, which is also believed to be causedby Marangoni flow.[9] Creeping of the slag on the refractory sample wall was apparent for allsamples, the height of the creeping layer being approximately same (15 mm), irrespective ofsample size.

Effect of addition of 5 wt% FeOx to the silicate slag was studied using a 30 mm sample underidentical experimental conditions as above. The gravimetric curves and tested refractory samples

of 18

are shown in Figure 3 in comparison with those for the slag without any FeOx. The addition of 5wt% FeOx significantly enhanced the penetration as evidenced by the prolonged net weight gainover two hours. The sectioned samples (after the test) clearly showed darker areas penetrated bythe iron oxide containing slag.

B. Calcium Ferrite Slag - Magnesia or Magchrome Refractories

The interactions of a magnesia and a magchrome refractory with a calcium ferrite type slag wereinvestigated. The chemical compositions of the refractories and the slag are shown in Table 2.The MgO refractory was a magnesite brick, which was different from the previous magnesiaspinel brick. Refractory samples were 30 mm in diameter and 44 mm in height. About 120 g slagsamples were used. The experiments were conducted at 1300 °C and an oxygen partial pressureof 10-6 atm, controlled using a CO2-CO mixture. The refractory-slag contact time was 1-2 hours.The gravimetric curves and the photographs of the samples before and after the test are shown inFigure 4.

The sectioned samples after the tests showed extensive penetration in both types of refractories,particularly the MgO refractory. This was also indicated by the bright slag-soaked sample surface(Figure 4). The penetration in the MgO refractory was so severe that the slag almost soaked theentire sample after 2 hours of contact. Both samples did not show any substantial slag linecorrosion. At the end of the tests, the slag composition was almost unchanged and the slag weightloss was, within the experimental uncertainty, equal to the sample weight gain. Therefore, thegravimetric data (after correction of wetting as discussed later) could be used directly as ameasure for the penetration of the slag. It is interesting to note that the penetration in themagchrome refractory was initially very fast but slowed dramatically after about 10 minutes andalmost ceased after about 15 minutes. The sectioned sample showed a slag penetration front atabout 32-34 mm from the hot face (slag-refractory contact).

C. Ladle Slags - Alumina Castable Refractories

Four alumina-based castable refractories were tested against a ladle type slag. The chemicalcompositions of the refractories and the slag are listed in Table 3. These alumina castables wereproduced by different manufacturers for steel ladle applications. Following the recommendedprocedures by the manufacturer, the castable mixes were cast into bricks and then fired.Cylindrical samples of 30 mm diameter were then core drilled from the fired bricks. The lengthof the four castables was different from one other (specified later in Figure 5) as some wereobtained from large precast blocks for use by rotary slag tests. The experiments were conductedat 1600 °C under argon and approximately 150 g of pre-melted slag was used. The slag-refractorycontact time was 2 hours. The gravimetric curves and the photographs of the tested refractorysamples are shown in Figure 5.

Considerable difference in slag attack was found among the four castable refractories. At the endof the experiments, the refractory samples had weight losses from 3 to 11 g and showedpenetration of the slag to a varying degree. Castables B and D were least penetrated (ordissolved) by the slag, and the weight losses were much less than the other two castables,showing a higher resistance to the slag. In comparison, Castables A and C were severely attacked

of 18

by the slag, due to extensive penetration and considerable dissolution/spalling of refractorygrains. It is interesting to note that the apparent attack on Castable A appeared to occur mainlythrough internal penetration while the slag attack on Castable C was more severe in the surfaceregion causing expansion and cracking. The gravimetric curves for Castable C showed that theslag attack occurred almost immediately after the refractory contacted the slag, while the rapidrise, and subsequent decrease in the gravimetric curve, was likely to be due to the cracking orspalling of the refractory grains.

IV. DISCUSSIONS

A. In-Situ Gravimetric Technique

The in-situ gravimetric curves relate to well defined physical processes as illustrated in Figure 6.The sharp weight increase at the initial slag-refractory contact was due to the wetting by the slagand the slag surface tension, exerting a “pull” on the cylindrical sample. The weight changes afterinitial contact showed the net effect of two simultaneous processes: slag penetration andrefractory dissolution. The weight gain in the early stage resulted from greater slag penetrationthan slag dissolution. The rate of penetration would decrease as the slag penetrated deeper, andthe weight gain would eventually be expected to reach a maximum at which the rate ofdissolution was equal to (and then exceeded) that of slag penetration.

The relative importance of slag penetration or dissolution processes, and indeed other radicalweight changes observed, depended on the slag-refractory system studied and varied from onesystem to another. For the magnesia spinel refractory and silicate slags (Figures 2 and 3), thegravimetric curves showed simultaneous penetration and dissolution similarly to that in Figure 6.For calcium ferrite slags in contact with the magnesia and magchrome refractories, thegravimetric curves showed fast and extensive penetration with little dissolution (Figure 4). In thecase of the ladle slag with four alumina castables, mixed behaviours were observed and severeattack and dissolution by ladle slags occurred in some cases (Figure 5).

A significant advantage of the new gravimetric technique is its capability to provide in-situkinetic information on the whole process of the slag-refractory interactions from the initialcontact, to severe slag attack and refractory failure. In particular, in-situ measurements providedirect insights into refractory wear due to fast and instantaneous reactions, such as cracking andspalling in some alumina castables when attacked by the ladle slag (Figure 5) and the rapidpenetration of calcium ferrite slags into the magnesia and magchrome refractories (Figure 4).Such information is difficult, if not impossible, to get from existing conventional techniquesbased on post-mortem analysis.

It is beyond the scope of the present publication to provide a comprehensive analysis of slag-refractory reactions and wear mechanisms in the foregoing three model systems. Detailedanalysis of gravimetric results obtained and the microscopic examination of the samples testedwill be the subject of future publications. The following discussions are focused on theinterpretation of gravimetric data in relation to the characteristics of wetting and slag penetration.

of 18

B. Wetting and Interfacial Properties

Figure 7 showed the gravimetric curves for the first 20 seconds after contact was made betweenthe magnesia spinel refractory and the silicate slag (Figures 2 and 3). Obviously, establishment ofinitial contact was very fast and this was immediately followed by very fast penetration, shownby a rapid weight increase for up to about 4 seconds, when the rate of weight gain then slowed.

The surface tension of the slags may be calculated from the initial weight increase (between free-hanging sample and slag wetted sample) if the non-smoothness of the porous refractory surfacewas ignored and initial fast penetration could be estimated. The first assumption was reasonableas the pores in the refractory were generally very small (median pore sizes around 10 µm) andliquid contact at such a surface would tend to smoothen out and form a continuous meniscussimilar to that formed at a smooth cylinder bottom. To correct the contribution from the initialfast penetration, the gravity and wetting force (estimated from the apparent weight) per unitcontact length of nominal circumference (F, dyn cm-1) measured at 4 sec was plotted as afunction of the nominal contact area (A, cm2) as shown in Figure 8. A linear dependence isobserved, within the experimental uncertainties, and expressed by the equation:

AF 35.385.423 += (1)

The second term represented the contribution of fast penetration which was proportional to thecontact area. Extrapolating to zero contact area gives an F value of 423 dyn cm-1. If the non-smoothness of the cylinder circumference is ignored, this should be equal to γ⋅cosθ where γ is thesurface tension of the slag and θ is the contact angle at the slag and refractory contact. The termγ⋅cosθ is an important measure of wetting characteristics and has been referred[10] to as“wettability parameter” or simply “wettability” (which will be used hereafter).

The surface tension of the silicate slag used has not been measured. However, extrapolation fromthe published surface tension data for similar systems[11,12] gives a value of γ ≈ 430 dyn cm-1. Nodata is available for the contact angle but it may be reasonably assumed to be small and hencecosθ should be close to unity. This leads to γ⋅cosθ ≈ 430 dyn cm-1, in close agreement with thatestimated from the gravimetric results. Figure 7 also showed the gravimetric curve for the slagwith addition of 5 wt% “FeOx”. Using Eq (1) to correct the initial fast penetration, the estimatedvalue for γ⋅cosθ was ∼ 429 dyn cm-1, slightly higher than that for the slag without iron oxide.

The gravimetric curves for the calcium ferrite slag with the magnesia and magchrome refractoriesin the first 20 seconds are presented in Figure 9. The initial weight increase was much higher thanthose observed for the silicate slags with the magnesia spinel refractory. The properties of thecalcium ferrite slags differed significantly from those of the silicate slags so Eq (1) was notapplicable. However, as there was little dissolution from both refractories, and the samplesretained their original cylinder shape at the end of the tests, the weight changes prior to, and afterthe slag separation from the refractory sample at the end of the tests, may be used to estimate thewettability. Loss of a substantial amount of the slag due to penetration caused a drop in the slagsurface and resulted in a slag film between the slag bath and the refractory sample before itsbreak-up. This was corrected for based on the known density of the slag.[13,14] The estimatedwettability values were 608 and 523 dyn cm-1 for magnesia and magchrome, respectively. This isbroadly consistent, assuming cosθ ≈ 1, with literature surface tension values[12] of 520-630 dyn

of 18

cm-1 for similar slags at 1350°C. The difference in estimated γ⋅cosθ between the two refractoriesappears to be likely due to the different wetting contacts rather than caused by the experimentaluncertainties. If we assume θ = 0° for the magnesia refractory, the slag contact angle at themagchrome surface would be approximately 30°.

C. Penetration of Slags

The penetration of the slags into the refractories is a complex process and is generally non-uniform in nature. The penetration front does not usually show as a clear-cut boundary, and veryoften rather as a blurred boundary region. In the areas which have been penetrated, not all poresare filled with the slag and the slag penetration through interconnected pores followed selectivechannels and pathways. This makes it difficult to quantify the true depth of penetration.Obviously any such depth should be treated as an average whether it is determined by the amount(weight) of the slag penetrated, or by virtual microscopic examination.

The present design of the slag-refractory contact is quite unique benefiting from the gravimetrictechnique which allows accurate control of the contact position. This helps to restrict the massflow across a well-defined contact area in one direction. In comparison, conventional techniquessuch as the “finger” test commonly use a partly “immersed’ sample[15] which complicates theconditions for data analysis and interpretation.

For refractory-slag systems without apparent slag creeping on the surface, the penetration of theslag could be estimated based on the gravimetric data after correction of the wettability and theweight of the slag film, if not negligible, at the slag-refractory contact. Applying to the calciumferrite slag with the magnesia and magchrome refractories, the amount of the penetrated slag wasestimated and is shown in Figure 10 as a function of contact time. Two short horizontal bars atthe end of each curve indicated the net weight gain by the refractory sample measured from thefree hanging sample after the slag-refractory contact was ceased. It is interesting to note that therate of penetration was almost same for both refractories during the first 13 minutes. As discussedpreviously, the wettability (γ⋅cosθ) for the magnesia refractory was likely to be higher than forthe magchrome, thus predicting a faster penetration of slag if other properties were identical.Lower wettability for the magchrome refractory, however, was compensated by a higher porosity(about 20%) and a relatively larger median pore size (∼ 10.5 µm) in comparison with themagnesia refractory (about 16% and 6.8 µm, respectively). The levelling off of penetration curveat longer period for the magchrome refractory was probably due to the reactions betweenpenetrated slag and refractory grains, causing changes in the composition and hence properties ofthe slag (such as melting temperature). Visual examination and chemical analysis of the sectionedsample indicated the penetration front to be as deep as 34 mm, which was higher than thatestimated (∼ 25 mm), if all pores had been filled by the slag. Apparently, some pores, particularlyin the areas close to the front, were unfilled.

Slag creeping provides possible alternative routes for slag to enter the refractory, and thuscomplicates the conditions. Creeping was observed when the silicate slag contacted the magnesiaspinel refractory. The height of creeping on the cylinder surface was about 15 mm for allrefractory samples of variable diameters. As the experiment using 5 wt% FeOx slag ran for 2hours, compared with 40 minutes for the slags without iron oxide, the observation of no

of 18

difference in the creeping height suggested that the creeping reached that height probably in arelatively short time. To estimate its contribution, the gravimetric data were analysed by plottingthe estimated net weight gain (after correction of wettability) per unit length of circumference orper unit contact area. For simplicity, a constant “wettability” was assumed and no correction wasmade for the slag film between the sample and the slag. The estimated net weight gain per unitlength of circumference differed considerably for three size samples, but the estimated weightgain per unit contact area was found to give reasonably close results as shown in Figure 11. Thissuggested that the mass flow due to penetration or dissolution was likely to be predominantlydetermined by the contact area, and the contribution from slag creeping was probably small. Theshort horizontal bars indicated the net weight gains measured from the free hanging samples afterthe tests, which were found to be higher than those estimated for the two smaller samples (Φ20and 30 mm) but significantly lower than that estimated for the large sample (Φ40 mm). Use of aconstant wettability and nominal circumference contact in the corrections was likely reasons forthe results observed with two small samples in Figure 11. The observation for the large sample ofΦ40 mm, however, could not be explained by this, or attributed to other experimentaluncertainties. The unexpected behaviour was likely to be caused by interference due to surfacetension, ie., when the gap between the sample and the crucible narrowed (the crucible wasslightly tapped towards the bottom).

For the addition of 5 wt% FeOx to the slag (Figure 3), the net weight gain was also estimated andshown in Figure 12. The estimated net weight gain was slightly higher than that measured at theend of the experiment, probably due to experimental uncertainties and the un-correctedcontribution from the slag film formed between the slag and the refractory sample. As the weightgain was the net effect of the penetration and dissolution, the increase in the penetration due to 5wt% FeOx addition was greater than that shown in Figure 12. The significantly enhanced (andprolonged) penetration likely resulted from increased wetting and possible preferential reactionsbetween iron oxide and periclase and spinel grains inside the refractory.

Also shown in Figure 12 are the estimated penetration (per contact area) of the calcium ferriteslag into the magnesia and magchrome refractories. Although the slags involved differedconsiderably, the results nevertheless clearly indicate the strong effect of iron oxide on thewetting and penetration of molten slags into magnesia based refractories. Significantly enhancedpenetration for the calcium ferrite slags was likely due to increased wettability (> 500 dyn cm-1

compared with < 430 dyn cm-1) and much lower kinematic viscosity (about 5 × 10-2 comparedwith 3 cm2 s-1).

D. Kinetics of Slag Penetration

Slag penetration in porous refractories is determined by a range of factors including interfacialproperties, slag chemistry, and structural characteristics of refractories. In principal, the drivingforce for slag penetration is due to wetting and capillary effects. The motion of liquid rise in acapillary of radius, r, is closely described by the following equation[16,17]

−= gh

rr

dtdhh ρθγ

ηcos2

8

2

(2)

of 18

where h is the height of liquid rise in the capillary, g the gravitational acceleration, η and ρ arethe viscosity and density of the liquid, respectively. For a horizontal capillary, the equationbecomes:

ηθγ

4cosr

dtdhh = (3)

and a simple solution could be readily derived as

trhη

θγ2cos= (4)

Eq (4) also applies to the initial capillary rise in a vertical tube over a short period (when h is verysmall and thus the second term in the bracket in Eq 2, ρgh, can be neglected).[18,19]

In order to apply these equations to liquid penetration in a porous solid, assumptions are oftenmade that the porous solid be treated as a bundle of infinitive parallel capillary tubes of the sameradius.[19,20] Apparently, this assumption is rarely valid in reality, as the three dimensional porestructure in porous refractories is extremely complex and, in many cases, this is furthercomplicated by reactions between penetrated slag and refractory grains.[16]

Taking the silicate slag and magnesia spinel refractory as an example, the depth of initialpenetration at 4 seconds could be estimated from Eq (1) to be about 0.83 mm, based on anapparent porosity of 18 % for the refractory, a density of 2.6 g cm-3 for the slag. However, thepenetration depth predicted by Eq (4) was 1.8 mm, based on a median pore radius of 3.5 µm(measured by mercury intrusion porisimeter), a wettability of γ⋅cosθ = 423 dyn cm-1, and the aviscosity of η = 8.7 poise.[21]

The difficulties in applying the simple models for capillary flow to complex porous media havebeen the subject of extensive studies. Various attempts have been made to introduce structuralparameters into these equations.[17,19,22,23] However, a satisfactory quantitative solution to theproblem is yet to be found and this remains an important subject for future research.

V. CONCLUSIONSA new gravimetric technique has been developed to provide direct information on theinstantaneous processes of dynamic wetting, penetration/dissolution, and wear of the refractoriesby molten slags. Such information can not be conveniently obtained from existing conventionaltechniques.

The gravimetric results of the three model slag-refractory systems showed complex behaviours ofthe slag-refractory interactions, from simultaneous penetration and dissolution, predominantpenetration with little dissolution, to severe penetration/cracking and refractory failure due to slagattack.

The slag penetration was generally non-uniform and could be very complex in some slag-refractory systems. The rate was determined by slag-refractory interfacial properties, slagproperties and chemical and structural characteristics of the refractories. The reactions between

of 18

the penetrated slag and the refractory grains played an important role as demonstrated bypenetration of the calcium ferrite type slag into the magchrome refractory.

Gravimetric data could be analysed to provide quantitative information on wetting andpenetration/dissolution of refractories by molten slags:

• Wettability, defined as γ⋅cosθ, could be estimated from the gravimetric data and thevalues derived were in good agreement with literature data.

• The amount of slag penetration could be readily calculated if the dissolution of therefractory was small, or alternatively, the net weight change due to simultaneouspenetration and dissolution could be estimated from the gravimetric data.

• Iron oxide in the slag had a significant effect on wetting and penetration of magnesiabased refractories, because of its influence on the wettability and the kinematic viscosityof the slags.

• The wettability of calcium ferrite slags on the magchrome refractory is poorer than on themagnesia refractory. Rates of slag penetration in the two refractories, however, are similardue to comparatively higher porosity and larger pore size in the magchrome refractory.

ACKNOWLEDGMENT

Financial support for this was provided by the Australian Government Cooperative ResearchCentres Program through the G K Williams Cooperative Research Centre for ExtractiveMetallurgy, a joint venture between the CSIRO Division of Minerals and the Department ofChemical Engineering, The University of Melbourne. The authors wish to thank Dr Colin Nexhipfor his help in early experimental trials.

Reference List

1. M. Kobayashi, M. Mishi and A. Miyamoto: Taikabutsu Overseas, 1982, 2(2), 5.

2. W. Lee and S. Zhang: International Materials Reviews, 1999, 44(3), 77.

3. W. E. Lee, P. Korgul, K. Goto and D. R. Wilson: Adv. Refract. Metall. Ind. II, Proc. Int.Symp., 1996, pp. 453-465.

4. K. Yamaguchi, F. Ogino and E. Kimura: EPD Congress, 1994, 803.

5. R. Crescent and M. Rigaud: Proc. Int. Symp. Adv. Refract. Metall. Ind., 1988, pp. 235-50.

6. K. Sugita: ISR'98, J. Wang et al, 1998, p. 9.

of 18

7. K. Mukai, J. M. Toguri, N. M. Stubina and J. Yoshitomi: ISIJ International, 1989, 29(6),469.

8. K. Mukai: Phil. Tran. Royal Soc., 1998, 356, 1015.

9. Z. Tao, K. Mukai and S. O. M. Yoshinaga: PacRim2, 1996, pp. 23-28.

10. P. R. Childambaram, G. R. Edwards and D. L. Olson: Metall. Trans., 1992, 23B, 215.

11. R. Benesch, R. Knihnicki and J. Janowski: Arch. Hutn., 1976, 21(4), 591.

12. V. D. Eisenhuttenleute: Slag Atlas, Verlag Stahleisen GmbH, Germany, 1995.

13. J. Henderson: Trans. AIME, 1964, 230, 501.

14. R. I. Gulyaeva, S. H. Shin, P. A. Kuznetsov and A. I. Okunev: Melts, 1990, 4(2), 80.

15. Y. Wanibe, S. Yokoyama, T. Itoh, T. Fujisawa and H. Sakao: Tetsu-To-Hagane, 1987,73(3), 491.

16. S. Zhang and A. Yamahuchi: J. Ceramic Soc. Japan, 1996, 104, 83.

17. Y. Wanibe, H. Tsuchida, T. Fujisawa and H. Sakao: Trans ISIJ, 1983, 23, 322.

18. J. Ligenza and R. B. Bernstein: J Am Chem Soc., 1951, 73, 4636.

19. K. Semlak and F. N. Rhines: Trans. AIME, 1958, 212, 325.

20. G. P. Martins, D. L. Olson and G. R. Edwards: Metall. Trans., 1988, 19B, 95.

21. J. S. Machin and T. B. Yee: J. Am. Ceram. Soc., 1954, 37, 177.

22. R. B. Bhagat and M. Singh: modeling of infiltration kinetics for the in-situ processing ofinorganic composites, in In-Situ Composites: Science and Technology, Minerals, Metalsand Materials Soc., 1994, p. 135.

23. Y. Wanibe, S. Yokoyama, T. Fujisawa and H. Sakao: Process Technology Proceedings,1986, 6, 911.

of 18

Gas in

��

SlagCrucible

Sample

Balance

PC��

��

Gas out

��

Fig. 1 Schematic diagram of experimental arrangement.

Samples after test

0

5

10

15

20

25

30

35

0:00 0:10 0:20 0:30 0:40 0:50Time, h:min

W, g

Φ30 mm

Φ20 mm

Φ40 mm

Fig. 2 Gravimetric curves for the magnesia spinel refractory in contact with the silicate slag at1500°C under argon and the refractory samples after the test.

of 18

Samples after test

5 wt% FeOxadded to slag

0

5

10

15

20

25

0:00 0:30 1:00 1:30 2:00 2:30Time, h:min

W, g

Silicate slag

+ 5%FeOx

Fig. 3 Gravimetric curves showing effect of adding 5 wt% FeOx on the penetration of the silicateslag into the magnesia spinel refractory (sample Φ30 mm, 1500°C, argon) and the refractorysamples after the test.

MgO

Before After

Magchrome0

10

20

30

40

50

0:00 0:30 1:00 1:30 2:00 2:30

Time, h:min

W, g

Magnesia

Magchrome

Fig. 4 Gravimetric curves for the magnesia and magchrome refractories in contact with thecalcium ferrite slag at 1300°C and pO2=10-6 atm (sample Φ30 mm) and the refractory samplesbefore and after the test.

of 18

A B C D

Samples after test

-12

-8

-4

0

4

8

12

16

0:00 0:30 1:00 1:30 2:00 2:30Time, h:min

W, g

Castable A

B D

C

D

C

BA

Fig. 5 Gravimetric curves showing interactions between the alumina castable refractories and theladle slag at 1600°C under argon atmosphere (refractory sample Φ30 mm) and the refractorysamples after the test (original sample length in mm, A = 52.3, B = 45.7, C=54, D=60.4).

0

2

4

6

8

10

12

0 2 4 6 8 10

Time

W

Initial contact

Fast penetration

P = D

D > P

P > D

P = PenetrationD = Dissolution

Fig. 6 Conceptual interpretation of a typical gravimetric curve.

of 18

0

5

10

15

0 5 10 15 20

Time, s

W, g

Φ20 mm

Φ30 mm

Φ40 mm

Φ30 mm, 5% FeOx

Fig. 7 Weight changes during initial contact between the magnesia spinel refractory and thesilicate slag.

y = 38.353x + 423.5

200

400

600

800

1000

0 5 10 15

Contact area, cm2

F, d

yn/c

m

Fig. 8 Effect of contact area on wetting force per unit circumference, F, at t = 4 sec contactbetween the magnesia spinel refractory and the silicate slag.

of 18

0

2

4

6

8

10

12

14

16

0 5 10 15 20

Time, s

W, g

Magnesia

Magchrome

Fig. 9 Weight change during initial contact between the calcium ferrite type slag and themagnesia and magchrome refractories.

0

5

10

15

20

25

30

0:00 0:30 1:00 1:30 2:00 2:30

Time, h:min

Slag

pen

etra

ted,

g

Magchrome

Magnesia

Fig. 10 Estimated amounts of the calcium ferrite slag penetrated into the magnesia andmagchrome refractories. The short horizontal bars show the actual weight gain measured fromthe free hanging refractory samples after the test.

of 18

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0:00 0:10 0:20 0:30 0:40 0:50Time, h:min

W/A

, g/c

m2

Φ30 mm

Φ20

Φ40 mm

Fig. 11 Estimated net weight gain per unit contact area for the magnesia spinel refractory incontact with the silicate slag at 1500 °C. The short horizontal bars show the measured net weightgain at the end of the test.

��

��

��

��

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

0:00 0:30 1:00 1:30 2:00 2:30

Time, h:min

W/A

, g/c

m2

0% FeOx

5% FeOx

Calcium ferrite slag75% FeOx

Silicate slag and magnesiaspinel refractory

Magnesia

Magchrome

Fig. 12 Effect of 5 wt% FeOx addition on the estimated net weight gain of the magnesia spinelrefractory in contact with the silicate slag at 1500 °C, in comparison with the estimatedpenetration of the calcium ferrite slag into the magnesia and magchrome refractories at 1300°Cand pO2=10-6 atm. The short horizontal bars show the measured net weight gain at the end of thetest.

of 18

Table 1 Compositions (wt%) of the silicate slag and the magnesia spinel refractory

SiO2 CaO Al2O3 MgOSlag 42 28 20 10

Refractory 0.2 0.9 6.5 92

Table 2 Compositions (wt%) of the calcium ferrite type slag and the magnesia andmagchrome refractories

Fe2O3 MgO Cr2O3 SiO2 CaO Al2O3 Cu2OSlag 75.7 19.4 5.3

MgO refractory 95 2.0 2.3 0.1Magchrome 52.5 23.55 1.42 0.71 12.6

Table 3. Compositions (wt%) of the ladle slag and the alumina castable refractories

Al2O3 CaO SiO2 MgO Fe2O3 MnOSlag 25.0 42.0 12.0 10.0 6.0 5.0

Castable A 85.4 2.1 2.2 8.5 0.5Castable B 91 0.6 2 3.2 0.01Castable C 92.8 1.7 0.1 4.9Castable D 93 0.3 3.2 0.2

ABSTRACTINTRODUCTIONEXPERIMENTALRESULTSSilicate Slag and Magnesia Spinel RefractoryCalcium Ferrite Slag - Magnesia or Magchrome RefractoriesLadle Slags - Alumina Castable Refractories

DISCUSSIONSIn-Situ Gravimetric TechniqueWetting and Interfacial PropertiesPenetration of SlagsKinetics of Slag Penetration

CONCLUSIONSACKNOWLEDGMENT

In-Situ Gravimetric Studies of Wetting, Penetration and Wear ......magnesia spinel brick (added...

Documents

Transcript of In-Situ Gravimetric Studies of Wetting, Penetration and Wear ......magnesia spinel brick (added...